/ ROUNDTABLE
Data science for embedded engineers
Avnet asked its own data scientist and machine learning experts what AI means to them and what engineers should consider when implementing AI and ML.
As a leader in its field, Avnet leverages technology to improve its own operations and customer service. In today’s business landscape, that means using artificial intelligence (AI) to improve the way we work. To learn more, this issue’s Roundtable includes Avnet Data Scientist Louis Arnaud and Digital Innovations Manager Samuele Scattolini as well as Monica Houston, manager of technical solutions engineering in Avnet’s Advanced Applications Group.
Is AI changing data science?
Refined data is the fuel powering the AI revolution. But data scientists have been capturing and analyzing data generated across all industries for decades. Before AI, this data was still being used to make business decisions. So how has AI changed data science?
“What has widely changed is the size of the dataset we are able to handle.”
More data leads to new ideas
While AI is changing the way we use data in a business environment, there must still be an underlying reason, a benefit, to using data to make business decisions. The introduction of AI and the enabling technologies around it creates new opportunities that need to be recognized before they can be grasped.
“The challenge, and what is exciting for me, is finding where the two worlds have a common layer.”
Using AI to generate new data
One of the things we’re told AI can do is relieve some of the pressure of work. If data scientists and embedded engineers rely so heavily on good data, can we now use AI to help clean that data or even create synthetic data for us?
“It's possible to create new data using generative AI.”
Check your expectations
There is a feeling within the data scientist community that expectations of AI will, for some time, remain too high. As Samuele Scattolini points out, the data used to train AI models will predominantly be refined to reflect the average. This means what comes out will also be average. It’s an interesting opinion, based on his experience of developing AI-enabled tools for Avnet.
“I recently heard a definition of generative AI that said ‘Now, mediocrity is free’.”
New ways to use data
Generative AI is receiving the majority of attention, mostly driven by large technology companies. But the data sets used for generative AI can be huge. This is a challenge for OEMs looking to implement the technology in constrained devices. But as Monica Houston points out, her experience demonstrates how quickly the technology is moving in this direction.
“Even a year ago I wouldn't have thought this could happen so quickly.”
A data scientist's advice for embedded engineers
When something appears to be intelligent, it is easy to forget that it is still fundamentally a machine that relies on inputs to create outputs. We asked Louis to offer embedded engineers like Monica some advice about handling data and using AI.
“Constantly question what you're receiving at the output.”